perf stat: Small random cleanups (v2)
Namhyung Kim
maintainercover perf stat: Small random cleanups (v2)
Hello,
I'm working on perf stat and I found some items to clean up. This time I removed runtime stats for per-thread aggregation mode which we can simply use thread map index to compare the shadow stat values in the rt_stat.
Changes from v1)
* renaming missing parts in headers (James)
* add a comment for map_idx (Ian)
* add Reviewed-by from James
The code is available at 'perf/stat-cleanup-v2' branch in
git://git.kernel.org/pub/scm/linux/kernel/git/namhyung/linux-perf.git
Thanks,
Namhyung
Namhyung Kim (7):
perf stat: Convert perf_stat_evsel.res_stats array perf stat: Don't call perf_stat_evsel_id_init() repeatedly perf stat: Rename saved_value->cpu_map_idx
perf stat: Use thread map index for shadow stat perf stat: Kill unused per-thread runtime stats perf stat: Don't compare runtime stat for shadow stats perf stat: Rename to aggr_cpu_id.thread_idx
tools/perf/builtin-stat.c | 54 ------ tools/perf/tests/topology.c | 10 +- tools/perf/util/cpumap.c | 8 +- tools/perf/util/cpumap.h | 2 +- tools/perf/util/stat-display.c | 32 ++-- tools/perf/util/stat-shadow.c | 320 ++++++++++++++++----------------- tools/perf/util/stat.c | 20 +-- tools/perf/util/stat.h | 10 +- 8 files changed, 189 insertions(+), 267 deletions(-)
base-commit: 62e64c9d2fd12839c02f1b3e8b873e7cb34e8720
Signature · 1 line
2.38.0.rc1.362.ged0d419d3c-goog
Namhyung Kim
maintainerPATCH 1/7 perf stat: Convert perf_stat_evsel.res_stats array
It uses only one member, no need to have it as an array.
--- tools/perf/util/stat-display.c | 2 +- tools/perf/util/stat.c | 10 +++------- tools/perf/util/stat.h | 2 +- 3 files changed, 5 insertions(+), 9 deletions(-)
tools/perf/util/stat-display.c +1 -1
@@ -67,7 +67,7 @@ static void print_noise(struct perf_stat_config *config,return ; ps = evsel->stats; -print_noise_pct (config,stddev_stats (&ps->res_stats[0 ]), avg); +print_noise_pct (config,stddev_stats (&ps->res_stats), avg); }static void print_cgroup (struct perf_stat_config *config,struct evsel *evsel)
tools/perf/util/stat.c +3 -7
@@ -132,12 +132,9 @@ static void perf_stat_evsel_id_init(struct evsel *evsel)static void evsel__reset_stat_priv (struct evsel *evsel) { -int i;struct perf_stat_evsel *ps = evsel->stats; -for (i =0 ; i <3 ; i++) -init_stats (&ps->res_stats[i]); - +init_stats (&ps->res_stats);perf_stat_evsel_id_init (evsel); }
@@ -440,7 +437,7 @@ int perf_stat_process_counter(struct perf_stat_config *config,struct perf_counts_values *aggr = &counter->counts->aggr;struct perf_stat_evsel *ps = counter->stats;u64 *count = counter->counts->aggr.values; -int i, ret; +int ret; aggr->val = aggr->ena = aggr->run =0 ;
@@ -458,8 +455,7 @@ int perf_stat_process_counter(struct perf_stat_config *config,evsel__compute_deltas (counter,-1 ,-1 , aggr);perf_counts_values__scale (aggr, config->scale, &counter->counts->scaled); -for (i =0 ; i <3 ; i++) -update_stats (&ps->res_stats[i], count[i]); +update_stats (&ps->res_stats, *count);if (verbose >0 ) { fprintf(config->output, "%s: %" PRIu64 " %" PRIu64 " %" PRIu64 "\n",
tools/perf/util/stat.h +1 -1
@@ -43,7 +43,7 @@ enum perf_stat_evsel_id { };struct perf_stat_evsel { -struct stats res_stats[3 ]; +struct stats res_stats;enum perf_stat_evsel_id id;u64 *group_data; };
Signature · 1 line
2.38.0.rc1.362.ged0d419d3c-goog
Namhyung Kim
maintainerPATCH 2/7 perf stat: Don't call perf_stat_evsel_id_init() repeatedly
In reply to Namhyung KimThe evsel__reset_stat_priv() is called more than once if user gave -r option for multiple run. But it doesn't need to re-initialize the id.
--- tools/perf/util/stat.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/util/stat.c +1 -1
@@ -135,7 +135,6 @@ static void evsel__reset_stat_priv(struct evsel *evsel)struct perf_stat_evsel *ps = evsel->stats;init_stats (&ps->res_stats); -perf_stat_evsel_id_init (evsel); }static int evsel__alloc_stat_priv (struct evsel *evsel)
@@ -143,6 +142,7 @@ static int evsel__alloc_stat_priv(struct evsel *evsel) evsel->stats =zalloc (sizeof (struct perf_stat_evsel ));if (evsel->stats ==NULL )return -ENOMEM ; +perf_stat_evsel_id_init (evsel);evsel__reset_stat_priv (evsel);return 0 ; }
Signature · 1 line
2.38.0.rc1.362.ged0d419d3c-goog
Namhyung Kim
maintainerPATCH 3/7 perf stat: Rename saved_value->cpu_map_idx
In reply to Namhyung KimThe cpu_map_idx fields is just to differentiate values from other entries. It doesn't need to be strictly cpu map index. Actually we can pass thread map index or aggr map index. So rename the fields first.
No functional change intended.
--- tools/perf/util/stat-shadow.c | 308 +++++++++++++++++----------------- tools/perf/util/stat.h | 6 +- 2 files changed, 157 insertions(+), 157 deletions(-)
tools/perf/util/stat-shadow.c +154 -154
@@ -33,7 +33,7 @@ struct saved_value {struct evsel *evsel;enum stat_type type;int ctx; -int cpu_map_idx; +int map_idx;/* cpu map index */ struct cgroup *cgrp;struct runtime_stat *stat;struct stats stats;
@@ -48,8 +48,8 @@ static int saved_value_cmp(struct rb_node *rb_node, const void *entry) rb_node);const struct saved_value *b = entry; -if (a->cpu_map_idx != b->cpu_map_idx) -return a->cpu_map_idx - b->cpu_map_idx; +if (a->map_idx != b->map_idx) +return a->map_idx - b->map_idx;/* * Previously the rbtree was used to link generic metrics.
@@ -106,7 +106,7 @@ static void saved_value_delete(struct rblist *rblist __maybe_unused, } static struct saved_value *saved_value_lookup(struct evsel *evsel, - int cpu_map_idx, + int map_idx, bool create, enum stat_type type, int ctx,
@@ -116,7 +116,7 @@ static struct saved_value *saved_value_lookup(struct evsel *evsel,struct rblist *rblist;struct rb_node *nd;struct saved_value dm = { - .cpu_map_idx = cpu_map_idx, + .map_idx = map_idx, .evsel = evsel, .type = type, .ctx = ctx,
@@ -215,10 +215,10 @@ struct runtime_stat_data {static void update_runtime_stat (struct runtime_stat *st,enum stat_type type, -int cpu_map_idx,u64 count, +int map_idx,u64 count,struct runtime_stat_data *rsd) { -struct saved_value *v =saved_value_lookup (NULL , cpu_map_idx, true, type, +struct saved_value *v =saved_value_lookup (NULL , map_idx, true, type, rsd->ctx, st, rsd->cgrp);if (v)
@@ -231,7 +231,7 @@ static void update_runtime_stat(struct runtime_stat *st,* instruction rates, etc: */ void perf_stat__update_shadow_stats (struct evsel *counter,u64 count, -int cpu_map_idx,struct runtime_stat *st) +int map_idx,struct runtime_stat *st) {u64 count_ns = count;struct saved_value *v;
@@ -243,88 +243,88 @@ void perf_stat__update_shadow_stats(struct evsel *counter, u64 count, count *= counter->scale;if (evsel__is_clock (counter)) -update_runtime_stat (st,STAT_NSECS , cpu_map_idx, count_ns, &rsd); +update_runtime_stat (st,STAT_NSECS , map_idx, count_ns, &rsd);else if (evsel__match (counter,HARDWARE ,HW_CPU_CYCLES )) -update_runtime_stat (st,STAT_CYCLES , cpu_map_idx, count, &rsd); +update_runtime_stat (st,STAT_CYCLES , map_idx, count, &rsd);else if (perf_stat_evsel__is (counter,CYCLES_IN_TX )) -update_runtime_stat (st,STAT_CYCLES_IN_TX , cpu_map_idx, count, &rsd); +update_runtime_stat (st,STAT_CYCLES_IN_TX , map_idx, count, &rsd);else if (perf_stat_evsel__is (counter,TRANSACTION_START )) -update_runtime_stat (st,STAT_TRANSACTION , cpu_map_idx, count, &rsd); +update_runtime_stat (st,STAT_TRANSACTION , map_idx, count, &rsd);else if (perf_stat_evsel__is (counter,ELISION_START )) -update_runtime_stat (st,STAT_ELISION , cpu_map_idx, count, &rsd); +update_runtime_stat (st,STAT_ELISION , map_idx, count, &rsd);else if (perf_stat_evsel__is (counter,TOPDOWN_TOTAL_SLOTS ))update_runtime_stat (st,STAT_TOPDOWN_TOTAL_SLOTS , - cpu_map_idx, count, &rsd); + map_idx, count, &rsd);else if (perf_stat_evsel__is (counter,TOPDOWN_SLOTS_ISSUED ))update_runtime_stat (st,STAT_TOPDOWN_SLOTS_ISSUED , - cpu_map_idx, count, &rsd); + map_idx, count, &rsd);else if (perf_stat_evsel__is (counter,TOPDOWN_SLOTS_RETIRED ))update_runtime_stat (st,STAT_TOPDOWN_SLOTS_RETIRED , - cpu_map_idx, count, &rsd); + map_idx, count, &rsd);else if (perf_stat_evsel__is (counter,TOPDOWN_FETCH_BUBBLES ))update_runtime_stat (st,STAT_TOPDOWN_FETCH_BUBBLES , - cpu_map_idx, count, &rsd); + map_idx, count, &rsd);else if (perf_stat_evsel__is (counter,TOPDOWN_RECOVERY_BUBBLES ))update_runtime_stat (st,STAT_TOPDOWN_RECOVERY_BUBBLES , - cpu_map_idx, count, &rsd); + map_idx, count, &rsd);else if (perf_stat_evsel__is (counter,TOPDOWN_RETIRING ))update_runtime_stat (st,STAT_TOPDOWN_RETIRING , - cpu_map_idx, count, &rsd); + map_idx, count, &rsd);else if (perf_stat_evsel__is (counter,TOPDOWN_BAD_SPEC ))update_runtime_stat (st,STAT_TOPDOWN_BAD_SPEC , - cpu_map_idx, count, &rsd); + map_idx, count, &rsd);else if (perf_stat_evsel__is (counter,TOPDOWN_FE_BOUND ))update_runtime_stat (st,STAT_TOPDOWN_FE_BOUND , - cpu_map_idx, count, &rsd); + map_idx, count, &rsd);else if (perf_stat_evsel__is (counter,TOPDOWN_BE_BOUND ))update_runtime_stat (st,STAT_TOPDOWN_BE_BOUND , - cpu_map_idx, count, &rsd); + map_idx, count, &rsd);else if (perf_stat_evsel__is (counter,TOPDOWN_HEAVY_OPS ))update_runtime_stat (st,STAT_TOPDOWN_HEAVY_OPS , - cpu_map_idx, count, &rsd); + map_idx, count, &rsd);else if (perf_stat_evsel__is (counter,TOPDOWN_BR_MISPREDICT ))update_runtime_stat (st,STAT_TOPDOWN_BR_MISPREDICT , - cpu_map_idx, count, &rsd); + map_idx, count, &rsd);else if (perf_stat_evsel__is (counter,TOPDOWN_FETCH_LAT ))update_runtime_stat (st,STAT_TOPDOWN_FETCH_LAT , - cpu_map_idx, count, &rsd); + map_idx, count, &rsd);else if (perf_stat_evsel__is (counter,TOPDOWN_MEM_BOUND ))update_runtime_stat (st,STAT_TOPDOWN_MEM_BOUND , - cpu_map_idx, count, &rsd); + map_idx, count, &rsd);else if (evsel__match (counter,HARDWARE ,HW_STALLED_CYCLES_FRONTEND ))update_runtime_stat (st,STAT_STALLED_CYCLES_FRONT , - cpu_map_idx, count, &rsd); + map_idx, count, &rsd);else if (evsel__match (counter,HARDWARE ,HW_STALLED_CYCLES_BACKEND ))update_runtime_stat (st,STAT_STALLED_CYCLES_BACK , - cpu_map_idx, count, &rsd); + map_idx, count, &rsd);else if (evsel__match (counter,HARDWARE ,HW_BRANCH_INSTRUCTIONS )) -update_runtime_stat (st,STAT_BRANCHES , cpu_map_idx, count, &rsd); +update_runtime_stat (st,STAT_BRANCHES , map_idx, count, &rsd);else if (evsel__match (counter,HARDWARE ,HW_CACHE_REFERENCES )) -update_runtime_stat (st,STAT_CACHEREFS , cpu_map_idx, count, &rsd); +update_runtime_stat (st,STAT_CACHEREFS , map_idx, count, &rsd);else if (evsel__match (counter,HW_CACHE ,HW_CACHE_L1D )) -update_runtime_stat (st,STAT_L1_DCACHE , cpu_map_idx, count, &rsd); +update_runtime_stat (st,STAT_L1_DCACHE , map_idx, count, &rsd);else if (evsel__match (counter,HW_CACHE ,HW_CACHE_L1I )) -update_runtime_stat (st,STAT_L1_ICACHE , cpu_map_idx, count, &rsd); +update_runtime_stat (st,STAT_L1_ICACHE , map_idx, count, &rsd);else if (evsel__match (counter,HW_CACHE ,HW_CACHE_LL )) -update_runtime_stat (st,STAT_LL_CACHE , cpu_map_idx, count, &rsd); +update_runtime_stat (st,STAT_LL_CACHE , map_idx, count, &rsd);else if (evsel__match (counter,HW_CACHE ,HW_CACHE_DTLB )) -update_runtime_stat (st,STAT_DTLB_CACHE , cpu_map_idx, count, &rsd); +update_runtime_stat (st,STAT_DTLB_CACHE , map_idx, count, &rsd);else if (evsel__match (counter,HW_CACHE ,HW_CACHE_ITLB )) -update_runtime_stat (st,STAT_ITLB_CACHE , cpu_map_idx, count, &rsd); +update_runtime_stat (st,STAT_ITLB_CACHE , map_idx, count, &rsd);else if (perf_stat_evsel__is (counter,SMI_NUM )) -update_runtime_stat (st,STAT_SMI_NUM , cpu_map_idx, count, &rsd); +update_runtime_stat (st,STAT_SMI_NUM , map_idx, count, &rsd);else if (perf_stat_evsel__is (counter,APERF )) -update_runtime_stat (st,STAT_APERF , cpu_map_idx, count, &rsd); +update_runtime_stat (st,STAT_APERF , map_idx, count, &rsd);if (counter->collect_stat) { - v =saved_value_lookup (counter, cpu_map_idx, true,STAT_NONE ,0 , st, + v =saved_value_lookup (counter, map_idx, true,STAT_NONE ,0 , st, rsd.cgrp);update_stats (&v->stats, count);if (counter->metric_leader) v->metric_total += count; }else if (counter->metric_leader) { v =saved_value_lookup (counter->metric_leader, - cpu_map_idx, true,STAT_NONE ,0 , st, rsd.cgrp); + map_idx, true,STAT_NONE ,0 , st, rsd.cgrp); v->metric_total += count; v->metric_other++; }
@@ -466,12 +466,12 @@ void perf_stat__collect_metric_expr(struct evlist *evsel_list) }static double runtime_stat_avg (struct runtime_stat *st, -enum stat_type type,int cpu_map_idx, +enum stat_type type,int map_idx,struct runtime_stat_data *rsd) {struct saved_value *v; - v =saved_value_lookup (NULL , cpu_map_idx, false, type, rsd->ctx, st, rsd->cgrp); + v =saved_value_lookup (NULL , map_idx, false, type, rsd->ctx, st, rsd->cgrp);if (!v)return 0.0 ;
@@ -479,12 +479,12 @@ static double runtime_stat_avg(struct runtime_stat *st, }static double runtime_stat_n (struct runtime_stat *st, -enum stat_type type,int cpu_map_idx, +enum stat_type type,int map_idx,struct runtime_stat_data *rsd) {struct saved_value *v; - v =saved_value_lookup (NULL , cpu_map_idx, false, type, rsd->ctx, st, rsd->cgrp); + v =saved_value_lookup (NULL , map_idx, false, type, rsd->ctx, st, rsd->cgrp);if (!v)return 0.0 ;
@@ -492,7 +492,7 @@ static double runtime_stat_n(struct runtime_stat *st, }static void print_stalled_cycles_frontend (struct perf_stat_config *config, -int cpu_map_idx,double avg, +int map_idx,double avg,struct perf_stat_output_ctx *out,struct runtime_stat *st,struct runtime_stat_data *rsd)
@@ -500,7 +500,7 @@ static void print_stalled_cycles_frontend(struct perf_stat_config *config,double total, ratio =0.0 ;const char *color; - total =runtime_stat_avg (st,STAT_CYCLES , cpu_map_idx, rsd); + total =runtime_stat_avg (st,STAT_CYCLES , map_idx, rsd);if (total) ratio = avg / total *100.0 ;
@@ -515,7 +515,7 @@ static void print_stalled_cycles_frontend(struct perf_stat_config *config, }static void print_stalled_cycles_backend (struct perf_stat_config *config, -int cpu_map_idx,double avg, +int map_idx,double avg,struct perf_stat_output_ctx *out,struct runtime_stat *st,struct runtime_stat_data *rsd)
@@ -523,7 +523,7 @@ static void print_stalled_cycles_backend(struct perf_stat_config *config,double total, ratio =0.0 ;const char *color; - total =runtime_stat_avg (st,STAT_CYCLES , cpu_map_idx, rsd); + total =runtime_stat_avg (st,STAT_CYCLES , map_idx, rsd);if (total) ratio = avg / total *100.0 ;
@@ -534,7 +534,7 @@ static void print_stalled_cycles_backend(struct perf_stat_config *config, }static void print_branch_misses (struct perf_stat_config *config, -int cpu_map_idx,double avg, +int map_idx,double avg,struct perf_stat_output_ctx *out,struct runtime_stat *st,struct runtime_stat_data *rsd)
@@ -542,7 +542,7 @@ static void print_branch_misses(struct perf_stat_config *config,double total, ratio =0.0 ;const char *color; - total =runtime_stat_avg (st,STAT_BRANCHES , cpu_map_idx, rsd); + total =runtime_stat_avg (st,STAT_BRANCHES , map_idx, rsd);if (total) ratio = avg / total *100.0 ;
@@ -553,7 +553,7 @@ static void print_branch_misses(struct perf_stat_config *config, }static void print_l1_dcache_misses (struct perf_stat_config *config, -int cpu_map_idx,double avg, +int map_idx,double avg,struct perf_stat_output_ctx *out,struct runtime_stat *st,struct runtime_stat_data *rsd)
@@ -561,7 +561,7 @@ static void print_l1_dcache_misses(struct perf_stat_config *config,double total, ratio =0.0 ;const char *color; - total =runtime_stat_avg (st,STAT_L1_DCACHE , cpu_map_idx, rsd); + total =runtime_stat_avg (st,STAT_L1_DCACHE , map_idx, rsd);if (total) ratio = avg / total *100.0 ;
@@ -572,7 +572,7 @@ static void print_l1_dcache_misses(struct perf_stat_config *config, }static void print_l1_icache_misses (struct perf_stat_config *config, -int cpu_map_idx,double avg, +int map_idx,double avg,struct perf_stat_output_ctx *out,struct runtime_stat *st,struct runtime_stat_data *rsd)
@@ -580,7 +580,7 @@ static void print_l1_icache_misses(struct perf_stat_config *config,double total, ratio =0.0 ;const char *color; - total =runtime_stat_avg (st,STAT_L1_ICACHE , cpu_map_idx, rsd); + total =runtime_stat_avg (st,STAT_L1_ICACHE , map_idx, rsd);if (total) ratio = avg / total *100.0 ;
@@ -590,7 +590,7 @@ static void print_l1_icache_misses(struct perf_stat_config *config, }static void print_dtlb_cache_misses (struct perf_stat_config *config, -int cpu_map_idx,double avg, +int map_idx,double avg,struct perf_stat_output_ctx *out,struct runtime_stat *st,struct runtime_stat_data *rsd)
@@ -598,7 +598,7 @@ static void print_dtlb_cache_misses(struct perf_stat_config *config,double total, ratio =0.0 ;const char *color; - total =runtime_stat_avg (st,STAT_DTLB_CACHE , cpu_map_idx, rsd); + total =runtime_stat_avg (st,STAT_DTLB_CACHE , map_idx, rsd);if (total) ratio = avg / total *100.0 ;
@@ -608,7 +608,7 @@ static void print_dtlb_cache_misses(struct perf_stat_config *config, }static void print_itlb_cache_misses (struct perf_stat_config *config, -int cpu_map_idx,double avg, +int map_idx,double avg,struct perf_stat_output_ctx *out,struct runtime_stat *st,struct runtime_stat_data *rsd)
@@ -616,7 +616,7 @@ static void print_itlb_cache_misses(struct perf_stat_config *config,double total, ratio =0.0 ;const char *color; - total =runtime_stat_avg (st,STAT_ITLB_CACHE , cpu_map_idx, rsd); + total =runtime_stat_avg (st,STAT_ITLB_CACHE , map_idx, rsd);if (total) ratio = avg / total *100.0 ;
@@ -626,7 +626,7 @@ static void print_itlb_cache_misses(struct perf_stat_config *config, }static void print_ll_cache_misses (struct perf_stat_config *config, -int cpu_map_idx,double avg, +int map_idx,double avg,struct perf_stat_output_ctx *out,struct runtime_stat *st,struct runtime_stat_data *rsd)
@@ -634,7 +634,7 @@ static void print_ll_cache_misses(struct perf_stat_config *config,double total, ratio =0.0 ;const char *color; - total =runtime_stat_avg (st,STAT_LL_CACHE , cpu_map_idx, rsd); + total =runtime_stat_avg (st,STAT_LL_CACHE , map_idx, rsd);if (total) ratio = avg / total *100.0 ;
@@ -692,61 +692,61 @@ static double sanitize_val(double x)return x; } -static double td_total_slots (int cpu_map_idx,struct runtime_stat *st, +static double td_total_slots (int map_idx,struct runtime_stat *st,struct runtime_stat_data *rsd) { -return runtime_stat_avg (st,STAT_TOPDOWN_TOTAL_SLOTS , cpu_map_idx, rsd); +return runtime_stat_avg (st,STAT_TOPDOWN_TOTAL_SLOTS , map_idx, rsd); } -static double td_bad_spec (int cpu_map_idx,struct runtime_stat *st, +static double td_bad_spec (int map_idx,struct runtime_stat *st,struct runtime_stat_data *rsd) {double bad_spec =0 ;double total_slots;double total; - total =runtime_stat_avg (st,STAT_TOPDOWN_SLOTS_ISSUED , cpu_map_idx, rsd) - -runtime_stat_avg (st,STAT_TOPDOWN_SLOTS_RETIRED , cpu_map_idx, rsd) + -runtime_stat_avg (st,STAT_TOPDOWN_RECOVERY_BUBBLES , cpu_map_idx, rsd); + total =runtime_stat_avg (st,STAT_TOPDOWN_SLOTS_ISSUED , map_idx, rsd) - +runtime_stat_avg (st,STAT_TOPDOWN_SLOTS_RETIRED , map_idx, rsd) + +runtime_stat_avg (st,STAT_TOPDOWN_RECOVERY_BUBBLES , map_idx, rsd); - total_slots =td_total_slots (cpu_map_idx, st, rsd); + total_slots =td_total_slots (map_idx, st, rsd);if (total_slots) bad_spec = total / total_slots;return sanitize_val (bad_spec); } -static double td_retiring (int cpu_map_idx,struct runtime_stat *st, +static double td_retiring (int map_idx,struct runtime_stat *st,struct runtime_stat_data *rsd) {double retiring =0 ; -double total_slots =td_total_slots (cpu_map_idx, st, rsd); +double total_slots =td_total_slots (map_idx, st, rsd);double ret_slots =runtime_stat_avg (st,STAT_TOPDOWN_SLOTS_RETIRED , - cpu_map_idx, rsd); + map_idx, rsd);if (total_slots) retiring = ret_slots / total_slots;return retiring; } -static double td_fe_bound (int cpu_map_idx,struct runtime_stat *st, +static double td_fe_bound (int map_idx,struct runtime_stat *st,struct runtime_stat_data *rsd) {double fe_bound =0 ; -double total_slots =td_total_slots (cpu_map_idx, st, rsd); +double total_slots =td_total_slots (map_idx, st, rsd);double fetch_bub =runtime_stat_avg (st,STAT_TOPDOWN_FETCH_BUBBLES , - cpu_map_idx, rsd); + map_idx, rsd);if (total_slots) fe_bound = fetch_bub / total_slots;return fe_bound; } -static double td_be_bound (int cpu_map_idx,struct runtime_stat *st, +static double td_be_bound (int map_idx,struct runtime_stat *st,struct runtime_stat_data *rsd) { -double sum = (td_fe_bound (cpu_map_idx, st, rsd) + -td_bad_spec (cpu_map_idx, st, rsd) + -td_retiring (cpu_map_idx, st, rsd)); +double sum = (td_fe_bound (map_idx, st, rsd) + +td_bad_spec (map_idx, st, rsd) + +td_retiring (map_idx, st, rsd));if (sum ==0 )return 0 ;return sanitize_val (1.0 - sum);
@@ -757,15 +757,15 @@ static double td_be_bound(int cpu_map_idx, struct runtime_stat *st,* the ratios we need to recreate the sum. */ -static double td_metric_ratio (int cpu_map_idx,enum stat_type type, +static double td_metric_ratio (int map_idx,enum stat_type type,struct runtime_stat *stat,struct runtime_stat_data *rsd) { -double sum =runtime_stat_avg (stat,STAT_TOPDOWN_RETIRING , cpu_map_idx, rsd) + -runtime_stat_avg (stat,STAT_TOPDOWN_FE_BOUND , cpu_map_idx, rsd) + -runtime_stat_avg (stat,STAT_TOPDOWN_BE_BOUND , cpu_map_idx, rsd) + -runtime_stat_avg (stat,STAT_TOPDOWN_BAD_SPEC , cpu_map_idx, rsd); -double d =runtime_stat_avg (stat, type, cpu_map_idx, rsd); +double sum =runtime_stat_avg (stat,STAT_TOPDOWN_RETIRING , map_idx, rsd) + +runtime_stat_avg (stat,STAT_TOPDOWN_FE_BOUND , map_idx, rsd) + +runtime_stat_avg (stat,STAT_TOPDOWN_BE_BOUND , map_idx, rsd) + +runtime_stat_avg (stat,STAT_TOPDOWN_BAD_SPEC , map_idx, rsd); +double d =runtime_stat_avg (stat, type, map_idx, rsd);if (sum)return d / sum;
@@ -777,23 +777,23 @@ static double td_metric_ratio(int cpu_map_idx, enum stat_type type,* We allow two missing. */ -static bool full_td (int cpu_map_idx,struct runtime_stat *stat, +static bool full_td (int map_idx,struct runtime_stat *stat,struct runtime_stat_data *rsd) {int c =0 ; -if (runtime_stat_avg (stat,STAT_TOPDOWN_RETIRING , cpu_map_idx, rsd) >0 ) +if (runtime_stat_avg (stat,STAT_TOPDOWN_RETIRING , map_idx, rsd) >0 ) c++; -if (runtime_stat_avg (stat,STAT_TOPDOWN_BE_BOUND , cpu_map_idx, rsd) >0 ) +if (runtime_stat_avg (stat,STAT_TOPDOWN_BE_BOUND , map_idx, rsd) >0 ) c++; -if (runtime_stat_avg (stat,STAT_TOPDOWN_FE_BOUND , cpu_map_idx, rsd) >0 ) +if (runtime_stat_avg (stat,STAT_TOPDOWN_FE_BOUND , map_idx, rsd) >0 ) c++; -if (runtime_stat_avg (stat,STAT_TOPDOWN_BAD_SPEC , cpu_map_idx, rsd) >0 ) +if (runtime_stat_avg (stat,STAT_TOPDOWN_BAD_SPEC , map_idx, rsd) >0 ) c++;return c >=2 ; } -static void print_smi_cost (struct perf_stat_config *config,int cpu_map_idx, +static void print_smi_cost (struct perf_stat_config *config,int map_idx,struct perf_stat_output_ctx *out,struct runtime_stat *st,struct runtime_stat_data *rsd)
@@ -801,9 +801,9 @@ static void print_smi_cost(struct perf_stat_config *config, int cpu_map_idx,double smi_num, aperf, cycles, cost =0.0 ;const char *color =NULL ; - smi_num =runtime_stat_avg (st,STAT_SMI_NUM , cpu_map_idx, rsd); - aperf =runtime_stat_avg (st,STAT_APERF , cpu_map_idx, rsd); - cycles =runtime_stat_avg (st,STAT_CYCLES , cpu_map_idx, rsd); + smi_num =runtime_stat_avg (st,STAT_SMI_NUM , map_idx, rsd); + aperf =runtime_stat_avg (st,STAT_APERF , map_idx, rsd); + cycles =runtime_stat_avg (st,STAT_CYCLES , map_idx, rsd);if ((cycles ==0 ) || (aperf ==0 ))return ;
@@ -820,7 +820,7 @@ static void print_smi_cost(struct perf_stat_config *config, int cpu_map_idx,static int prepare_metric (struct evsel **metric_events,struct metric_ref *metric_refs,struct expr_parse_ctx *pctx, -int cpu_map_idx, +int map_idx,struct runtime_stat *st) {double scale;
@@ -859,7 +859,7 @@ static int prepare_metric(struct evsel **metric_events,abort (); } }else { - v =saved_value_lookup (metric_events[i], cpu_map_idx, false, + v =saved_value_lookup (metric_events[i], map_idx, false,STAT_NONE ,0 , st, metric_events[i]->cgrp);if (!v)
@@ -897,7 +897,7 @@ static void generic_metric(struct perf_stat_config *config,const char *metric_name, const char *metric_unit, int runtime, - int cpu_map_idx, + int map_idx, struct perf_stat_output_ctx *out, struct runtime_stat *st) {
@@ -915,7 +915,7 @@ static void generic_metric(struct perf_stat_config *config, pctx->sctx.user_requested_cpu_list =strdup (config->user_requested_cpu_list); pctx->sctx.runtime = runtime; pctx->sctx.system_wide = config->system_wide; - i =prepare_metric (metric_events, metric_refs, pctx, cpu_map_idx, st); + i =prepare_metric (metric_events, metric_refs, pctx, map_idx, st);if (i <0 ) {expr__ctx_free (pctx);return ;
@@ -960,7 +960,7 @@ static void generic_metric(struct perf_stat_config *config,expr__ctx_free (pctx); } -double test_generic_metric (struct metric_expr *mexp,int cpu_map_idx,struct runtime_stat *st) +double test_generic_metric (struct metric_expr *mexp,int map_idx,struct runtime_stat *st) {struct expr_parse_ctx *pctx;double ratio =0.0 ;
@@ -969,7 +969,7 @@ double test_generic_metric(struct metric_expr *mexp, int cpu_map_idx, struct runif (!pctx)return NAN ; -if (prepare_metric (mexp->metric_events, mexp->metric_refs, pctx, cpu_map_idx, st) <0 ) +if (prepare_metric (mexp->metric_events, mexp->metric_refs, pctx, map_idx, st) <0 ) goto out;if (expr__parse (&ratio, pctx, mexp->metric_expr))
@@ -982,7 +982,7 @@ double test_generic_metric(struct metric_expr *mexp, int cpu_map_idx, struct runvoid perf_stat__print_shadow_stats (struct perf_stat_config *config,struct evsel *evsel, -double avg,int cpu_map_idx, +double avg,int map_idx,struct perf_stat_output_ctx *out,struct rblist *metric_events,struct runtime_stat *st)
@@ -1001,7 +1001,7 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config,if (config->iostat_run) {iostat_print_metric (config, evsel, out); }else if (evsel__match (evsel,HARDWARE ,HW_INSTRUCTIONS )) { - total =runtime_stat_avg (st,STAT_CYCLES , cpu_map_idx, &rsd); + total =runtime_stat_avg (st,STAT_CYCLES , map_idx, &rsd);if (total) { ratio = avg / total;
@@ -1011,11 +1011,11 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config,print_metric (config, ctxp,NULL ,NULL ,"insn per cycle" ,0 ); } - total =runtime_stat_avg (st,STAT_STALLED_CYCLES_FRONT , cpu_map_idx, &rsd); + total =runtime_stat_avg (st,STAT_STALLED_CYCLES_FRONT , map_idx, &rsd); total =max (total,runtime_stat_avg (st,STAT_STALLED_CYCLES_BACK , - cpu_map_idx, &rsd)); + map_idx, &rsd));if (total && avg) { out->new_line (config, ctxp);
@@ -1025,8 +1025,8 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, ratio); } }else if (evsel__match (evsel,HARDWARE ,HW_BRANCH_MISSES )) { -if (runtime_stat_n (st,STAT_BRANCHES , cpu_map_idx, &rsd) !=0 ) -print_branch_misses (config, cpu_map_idx, avg, out, st, &rsd); +if (runtime_stat_n (st,STAT_BRANCHES , map_idx, &rsd) !=0 ) +print_branch_misses (config, map_idx, avg, out, st, &rsd);else print_metric (config, ctxp,NULL ,NULL ,"of all branches" ,0 ); } else if (
@@ -1035,8 +1035,8 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, ((PERF_COUNT_HW_CACHE_OP_READ) << 8) | ((PERF_COUNT_HW_CACHE_RESULT_MISS) << 16))) { -if (runtime_stat_n (st,STAT_L1_DCACHE , cpu_map_idx, &rsd) !=0 ) -print_l1_dcache_misses (config, cpu_map_idx, avg, out, st, &rsd); +if (runtime_stat_n (st,STAT_L1_DCACHE , map_idx, &rsd) !=0 ) +print_l1_dcache_misses (config, map_idx, avg, out, st, &rsd);else print_metric (config, ctxp,NULL ,NULL ,"of all L1-dcache accesses" ,0 ); } else if (
@@ -1045,8 +1045,8 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, ((PERF_COUNT_HW_CACHE_OP_READ) << 8) | ((PERF_COUNT_HW_CACHE_RESULT_MISS) << 16))) { -if (runtime_stat_n (st,STAT_L1_ICACHE , cpu_map_idx, &rsd) !=0 ) -print_l1_icache_misses (config, cpu_map_idx, avg, out, st, &rsd); +if (runtime_stat_n (st,STAT_L1_ICACHE , map_idx, &rsd) !=0 ) +print_l1_icache_misses (config, map_idx, avg, out, st, &rsd);else print_metric (config, ctxp,NULL ,NULL ,"of all L1-icache accesses" ,0 ); } else if (
@@ -1055,8 +1055,8 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, ((PERF_COUNT_HW_CACHE_OP_READ) << 8) | ((PERF_COUNT_HW_CACHE_RESULT_MISS) << 16))) { -if (runtime_stat_n (st,STAT_DTLB_CACHE , cpu_map_idx, &rsd) !=0 ) -print_dtlb_cache_misses (config, cpu_map_idx, avg, out, st, &rsd); +if (runtime_stat_n (st,STAT_DTLB_CACHE , map_idx, &rsd) !=0 ) +print_dtlb_cache_misses (config, map_idx, avg, out, st, &rsd);else print_metric (config, ctxp,NULL ,NULL ,"of all dTLB cache accesses" ,0 ); } else if (
@@ -1065,8 +1065,8 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, ((PERF_COUNT_HW_CACHE_OP_READ) << 8) | ((PERF_COUNT_HW_CACHE_RESULT_MISS) << 16))) { -if (runtime_stat_n (st,STAT_ITLB_CACHE , cpu_map_idx, &rsd) !=0 ) -print_itlb_cache_misses (config, cpu_map_idx, avg, out, st, &rsd); +if (runtime_stat_n (st,STAT_ITLB_CACHE , map_idx, &rsd) !=0 ) +print_itlb_cache_misses (config, map_idx, avg, out, st, &rsd);else print_metric (config, ctxp,NULL ,NULL ,"of all iTLB cache accesses" ,0 ); } else if (
@@ -1075,27 +1075,27 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, ((PERF_COUNT_HW_CACHE_OP_READ) << 8) | ((PERF_COUNT_HW_CACHE_RESULT_MISS) << 16))) { -if (runtime_stat_n (st,STAT_LL_CACHE , cpu_map_idx, &rsd) !=0 ) -print_ll_cache_misses (config, cpu_map_idx, avg, out, st, &rsd); +if (runtime_stat_n (st,STAT_LL_CACHE , map_idx, &rsd) !=0 ) +print_ll_cache_misses (config, map_idx, avg, out, st, &rsd);else print_metric (config, ctxp,NULL ,NULL ,"of all LL-cache accesses" ,0 ); } else if (evsel__match(evsel, HARDWARE, HW_CACHE_MISSES)) { - total =runtime_stat_avg (st,STAT_CACHEREFS , cpu_map_idx, &rsd); + total =runtime_stat_avg (st,STAT_CACHEREFS , map_idx, &rsd);if (total) ratio = avg *100 / total; -if (runtime_stat_n (st,STAT_CACHEREFS , cpu_map_idx, &rsd) !=0 ) +if (runtime_stat_n (st,STAT_CACHEREFS , map_idx, &rsd) !=0 )print_metric (config, ctxp,NULL ,"%8.3f %%" ,"of all cache refs" , ratio);else print_metric (config, ctxp,NULL ,NULL ,"of all cache refs" ,0 ); } else if (evsel__match(evsel, HARDWARE, HW_STALLED_CYCLES_FRONTEND)) { -print_stalled_cycles_frontend (config, cpu_map_idx, avg, out, st, &rsd); +print_stalled_cycles_frontend (config, map_idx, avg, out, st, &rsd); } else if (evsel__match(evsel, HARDWARE, HW_STALLED_CYCLES_BACKEND)) { -print_stalled_cycles_backend (config, cpu_map_idx, avg, out, st, &rsd); +print_stalled_cycles_backend (config, map_idx, avg, out, st, &rsd); } else if (evsel__match(evsel, HARDWARE, HW_CPU_CYCLES)) { - total =runtime_stat_avg (st,STAT_NSECS , cpu_map_idx, &rsd); + total =runtime_stat_avg (st,STAT_NSECS , map_idx, &rsd);if (total) { ratio = avg / total;
@@ -1104,7 +1104,7 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config,print_metric (config, ctxp,NULL ,NULL ,"Ghz" ,0 ); } }else if (perf_stat_evsel__is (evsel,CYCLES_IN_TX )) { - total =runtime_stat_avg (st,STAT_CYCLES , cpu_map_idx, &rsd); + total =runtime_stat_avg (st,STAT_CYCLES , map_idx, &rsd);if (total) print_metric(config, ctxp, NULL,
@@ -1114,8 +1114,8 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config,print_metric (config, ctxp,NULL ,NULL ,"transactional cycles" ,0 ); }else if (perf_stat_evsel__is (evsel,CYCLES_IN_TX_CP )) { - total =runtime_stat_avg (st,STAT_CYCLES , cpu_map_idx, &rsd); - total2 =runtime_stat_avg (st,STAT_CYCLES_IN_TX , cpu_map_idx, &rsd); + total =runtime_stat_avg (st,STAT_CYCLES , map_idx, &rsd); + total2 =runtime_stat_avg (st,STAT_CYCLES_IN_TX , map_idx, &rsd);if (total2 < avg) total2 = avg;
@@ -1125,19 +1125,19 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config,else print_metric (config, ctxp,NULL ,NULL ,"aborted cycles" ,0 ); }else if (perf_stat_evsel__is (evsel,TRANSACTION_START )) { - total =runtime_stat_avg (st,STAT_CYCLES_IN_TX , cpu_map_idx, &rsd); + total =runtime_stat_avg (st,STAT_CYCLES_IN_TX , map_idx, &rsd);if (avg) ratio = total / avg; -if (runtime_stat_n (st,STAT_CYCLES_IN_TX , cpu_map_idx, &rsd) !=0 ) +if (runtime_stat_n (st,STAT_CYCLES_IN_TX , map_idx, &rsd) !=0 )print_metric (config, ctxp,NULL ,"%8.0f" ,"cycles / transaction" , ratio);else print_metric (config, ctxp,NULL ,NULL ,"cycles / transaction" ,0 ); }else if (perf_stat_evsel__is (evsel,ELISION_START )) { - total =runtime_stat_avg (st,STAT_CYCLES_IN_TX , cpu_map_idx, &rsd); + total =runtime_stat_avg (st,STAT_CYCLES_IN_TX , map_idx, &rsd);if (avg) ratio = total / avg;
@@ -1150,28 +1150,28 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config,else print_metric (config, ctxp,NULL ,NULL ,"CPUs utilized" ,0 ); }else if (perf_stat_evsel__is (evsel,TOPDOWN_FETCH_BUBBLES )) { -double fe_bound =td_fe_bound (cpu_map_idx, st, &rsd); +double fe_bound =td_fe_bound (map_idx, st, &rsd);if (fe_bound >0.2 ) color =PERF_COLOR_RED ;print_metric (config, ctxp, color,"%8.1f%%" ,"frontend bound" , fe_bound *100. ); }else if (perf_stat_evsel__is (evsel,TOPDOWN_SLOTS_RETIRED )) { -double retiring =td_retiring (cpu_map_idx, st, &rsd); +double retiring =td_retiring (map_idx, st, &rsd);if (retiring >0.7 ) color =PERF_COLOR_GREEN ;print_metric (config, ctxp, color,"%8.1f%%" ,"retiring" , retiring *100. ); }else if (perf_stat_evsel__is (evsel,TOPDOWN_RECOVERY_BUBBLES )) { -double bad_spec =td_bad_spec (cpu_map_idx, st, &rsd); +double bad_spec =td_bad_spec (map_idx, st, &rsd);if (bad_spec >0.1 ) color =PERF_COLOR_RED ;print_metric (config, ctxp, color,"%8.1f%%" ,"bad speculation" , bad_spec *100. ); }else if (perf_stat_evsel__is (evsel,TOPDOWN_SLOTS_ISSUED )) { -double be_bound =td_be_bound (cpu_map_idx, st, &rsd); +double be_bound =td_be_bound (map_idx, st, &rsd);const char *name ="backend bound" ;static int have_recovery_bubbles =-1 ;
@@ -1184,14 +1184,14 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config,if (be_bound >0.2 ) color =PERF_COLOR_RED ; -if (td_total_slots (cpu_map_idx, st, &rsd) >0 ) +if (td_total_slots (map_idx, st, &rsd) >0 )print_metric (config, ctxp, color,"%8.1f%%" , name, be_bound *100. );else print_metric (config, ctxp,NULL ,NULL , name,0 ); }else if (perf_stat_evsel__is (evsel,TOPDOWN_RETIRING ) && -full_td (cpu_map_idx, st, &rsd)) { -double retiring =td_metric_ratio (cpu_map_idx, +full_td (map_idx, st, &rsd)) { +double retiring =td_metric_ratio (map_idx,STAT_TOPDOWN_RETIRING , st, &rsd);if (retiring >0.7 )
@@ -1199,8 +1199,8 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config,print_metric (config, ctxp, color,"%8.1f%%" ,"Retiring" , retiring *100. ); }else if (perf_stat_evsel__is (evsel,TOPDOWN_FE_BOUND ) && -full_td (cpu_map_idx, st, &rsd)) { -double fe_bound =td_metric_ratio (cpu_map_idx, +full_td (map_idx, st, &rsd)) { +double fe_bound =td_metric_ratio (map_idx,STAT_TOPDOWN_FE_BOUND , st, &rsd);if (fe_bound >0.2 )
@@ -1208,8 +1208,8 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config,print_metric (config, ctxp, color,"%8.1f%%" ,"Frontend Bound" , fe_bound *100. ); }else if (perf_stat_evsel__is (evsel,TOPDOWN_BE_BOUND ) && -full_td (cpu_map_idx, st, &rsd)) { -double be_bound =td_metric_ratio (cpu_map_idx, +full_td (map_idx, st, &rsd)) { +double be_bound =td_metric_ratio (map_idx,STAT_TOPDOWN_BE_BOUND , st, &rsd);if (be_bound >0.2 )
@@ -1217,8 +1217,8 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config,print_metric (config, ctxp, color,"%8.1f%%" ,"Backend Bound" , be_bound *100. ); }else if (perf_stat_evsel__is (evsel,TOPDOWN_BAD_SPEC ) && -full_td (cpu_map_idx, st, &rsd)) { -double bad_spec =td_metric_ratio (cpu_map_idx, +full_td (map_idx, st, &rsd)) { +double bad_spec =td_metric_ratio (map_idx,STAT_TOPDOWN_BAD_SPEC , st, &rsd);if (bad_spec >0.1 )
@@ -1226,11 +1226,11 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config,print_metric (config, ctxp, color,"%8.1f%%" ,"Bad Speculation" , bad_spec *100. ); }else if (perf_stat_evsel__is (evsel,TOPDOWN_HEAVY_OPS ) && -full_td (cpu_map_idx, st, &rsd) && (config->topdown_level >1 )) { -double retiring =td_metric_ratio (cpu_map_idx, +full_td (map_idx, st, &rsd) && (config->topdown_level >1 )) { +double retiring =td_metric_ratio (map_idx,STAT_TOPDOWN_RETIRING , st, &rsd); -double heavy_ops =td_metric_ratio (cpu_map_idx, +double heavy_ops =td_metric_ratio (map_idx,STAT_TOPDOWN_HEAVY_OPS , st, &rsd);double light_ops = retiring - heavy_ops;
@@ -1246,11 +1246,11 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config,print_metric (config, ctxp, color,"%8.1f%%" ,"Light Operations" , light_ops *100. ); }else if (perf_stat_evsel__is (evsel,TOPDOWN_BR_MISPREDICT ) && -full_td (cpu_map_idx, st, &rsd) && (config->topdown_level >1 )) { -double bad_spec =td_metric_ratio (cpu_map_idx, +full_td (map_idx, st, &rsd) && (config->topdown_level >1 )) { +double bad_spec =td_metric_ratio (map_idx,STAT_TOPDOWN_BAD_SPEC , st, &rsd); -double br_mis =td_metric_ratio (cpu_map_idx, +double br_mis =td_metric_ratio (map_idx,STAT_TOPDOWN_BR_MISPREDICT , st, &rsd);double m_clears = bad_spec - br_mis;
@@ -1266,11 +1266,11 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config,print_metric (config, ctxp, color,"%8.1f%%" ,"Machine Clears" , m_clears *100. ); }else if (perf_stat_evsel__is (evsel,TOPDOWN_FETCH_LAT ) && -full_td (cpu_map_idx, st, &rsd) && (config->topdown_level >1 )) { -double fe_bound =td_metric_ratio (cpu_map_idx, +full_td (map_idx, st, &rsd) && (config->topdown_level >1 )) { +double fe_bound =td_metric_ratio (map_idx,STAT_TOPDOWN_FE_BOUND , st, &rsd); -double fetch_lat =td_metric_ratio (cpu_map_idx, +double fetch_lat =td_metric_ratio (map_idx,STAT_TOPDOWN_FETCH_LAT , st, &rsd);double fetch_bw = fe_bound - fetch_lat;
@@ -1286,11 +1286,11 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config,print_metric (config, ctxp, color,"%8.1f%%" ,"Fetch Bandwidth" , fetch_bw *100. ); }else if (perf_stat_evsel__is (evsel,TOPDOWN_MEM_BOUND ) && -full_td (cpu_map_idx, st, &rsd) && (config->topdown_level >1 )) { -double be_bound =td_metric_ratio (cpu_map_idx, +full_td (map_idx, st, &rsd) && (config->topdown_level >1 )) { +double be_bound =td_metric_ratio (map_idx,STAT_TOPDOWN_BE_BOUND , st, &rsd); -double mem_bound =td_metric_ratio (cpu_map_idx, +double mem_bound =td_metric_ratio (map_idx,STAT_TOPDOWN_MEM_BOUND , st, &rsd);double core_bound = be_bound - mem_bound;
@@ -1308,12 +1308,12 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, }else if (evsel->metric_expr) {generic_metric (config, evsel->metric_expr, evsel->metric_events,NULL , evsel->name, evsel->metric_name,NULL ,1 , - cpu_map_idx, out, st); - }else if (runtime_stat_n (st,STAT_NSECS , cpu_map_idx, &rsd) !=0 ) { + map_idx, out, st); + }else if (runtime_stat_n (st,STAT_NSECS , map_idx, &rsd) !=0 ) {char unit =' ' ;char unit_buf[10 ] ="/sec" ; - total =runtime_stat_avg (st,STAT_NSECS , cpu_map_idx, &rsd); + total =runtime_stat_avg (st,STAT_NSECS , map_idx, &rsd);if (total) ratio =convert_unit_double (1000000000.0 * avg / total, &unit);
@@ -1321,7 +1321,7 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config,snprintf (unit_buf,sizeof (unit_buf),"%c/sec" , unit);print_metric (config, ctxp,NULL ,"%8.3f" , unit_buf, ratio); }else if (perf_stat_evsel__is (evsel,SMI_NUM )) { -print_smi_cost (config, cpu_map_idx, out, st, &rsd); +print_smi_cost (config, map_idx, out, st, &rsd); }else { num =0 ; }
@@ -1335,7 +1335,7 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config,generic_metric (config, mexp->metric_expr, mexp->metric_events, mexp->metric_refs, evsel->name, mexp->metric_name, mexp->metric_unit, mexp->runtime, - cpu_map_idx, out, st); + map_idx, out, st); } }if (num ==0 )
tools/perf/util/stat.h +3 -3
@@ -234,7 +234,7 @@ void perf_stat__init_shadow_stats(void);void perf_stat__reset_shadow_stats (void );void perf_stat__reset_shadow_per_stat (struct runtime_stat *st);void perf_stat__update_shadow_stats (struct evsel *counter,u64 count, -int cpu_map_idx,struct runtime_stat *st); +int map_idx,struct runtime_stat *st);struct perf_stat_output_ctx {void *ctx;print_metric_t print_metric;
@@ -244,7 +244,7 @@ struct perf_stat_output_ctx {void perf_stat__print_shadow_stats (struct perf_stat_config *config,struct evsel *evsel, -double avg,int cpu, +double avg,int map_idx,struct perf_stat_output_ctx *out,struct rblist *metric_events,struct runtime_stat *st);
@@ -279,5 +279,5 @@ void evlist__print_counters(struct evlist *evlist, struct perf_stat_config *conf struct target *_target, struct timespec *ts, int argc, const char **argv);struct metric_expr ; -double test_generic_metric (struct metric_expr *mexp,int cpu_map_idx,struct runtime_stat *st); +double test_generic_metric (struct metric_expr *mexp,int map_idx,struct runtime_stat *st);#endif
Signature · 1 line
2.38.0.rc1.362.ged0d419d3c-goog
Namhyung Kim
maintainerPATCH 4/7 perf stat: Use thread map index for shadow stat
In reply to Namhyung KimWhen AGGR_THREAD is active, it aggregates the values for each thread. Previously it used cpu map index which is invalid for AGGR_THREAD so it had to use separate runtime stats with index 0.
But it can just use the rt_stat with thread_map_index. Rename the first_shadow_map_idx() and make it return the thread index.
--- tools/perf/util/stat-display.c | 20 +++++++++----------- tools/perf/util/stat-shadow.c | 2 +- tools/perf/util/stat.c | 8 ++------ 3 files changed, 12 insertions(+), 18 deletions(-)
tools/perf/util/stat-display.c +9 -11
@@ -442,7 +442,7 @@ static void print_metric_header(struct perf_stat_config *config,fprintf (os->fh,"%*s " , config->metric_only_len, unit); } -static int first_shadow_cpu_map_idx (struct perf_stat_config *config, +static int first_shadow_map_idx (struct perf_stat_config *config,struct evsel *evsel,const struct aggr_cpu_id *id) {struct perf_cpu_map *cpus =evsel__cpus (evsel);
@@ -452,6 +452,9 @@ static int first_shadow_cpu_map_idx(struct perf_stat_config *config,if (config->aggr_mode ==AGGR_NONE )return perf_cpu_map__idx (cpus, id->cpu); +if (config->aggr_mode ==AGGR_THREAD ) +return id->thread; +if (!config->aggr_get_id)return 0 ;
@@ -646,7 +649,7 @@ static void printout(struct perf_stat_config *config, struct aggr_cpu_id id, int }perf_stat__print_shadow_stats (config, counter, uval, -first_shadow_cpu_map_idx (config, counter, &id), +first_shadow_map_idx (config, counter, &id), &out, &config->metric_events, st);if (!config->csv_output && !config->metric_only && !config->json_output) {print_noise (config, counter, noise);
@@ -676,7 +679,7 @@ static void aggr_update_shadow(struct perf_stat_config *config, val +=perf_counts (counter->counts, idx,0 )->val; }perf_stat__update_shadow_stats (counter, val, -first_shadow_cpu_map_idx (config, counter, &id), +first_shadow_map_idx (config, counter, &id), &rt_stat); } }
@@ -979,14 +982,9 @@ static void print_aggr_thread(struct perf_stat_config *config,fprintf (output,"%s" , prefix); id = buf[thread].id; -if (config->stats) -printout (config, id,0 , buf[thread].counter, buf[thread].uval, - prefix, buf[thread].run, buf[thread].ena,1.0 , - &config->stats[id.thread]); -else -printout (config, id,0 , buf[thread].counter, buf[thread].uval, - prefix, buf[thread].run, buf[thread].ena,1.0 , - &rt_stat); +printout (config, id,0 , buf[thread].counter, buf[thread].uval, + prefix, buf[thread].run, buf[thread].ena,1.0 , + &rt_stat);fputc ('\n' , output); }
tools/perf/util/stat-shadow.c +1 -1
@@ -33,7 +33,7 @@ struct saved_value {struct evsel *evsel;enum stat_type type;int ctx; -int map_idx;/* cpu map index */ +int map_idx;/* cpu or thread map index */ struct cgroup *cgrp;struct runtime_stat *stat;struct stats stats;
tools/perf/util/stat.c +2 -6
@@ -389,12 +389,8 @@ process_counter_values(struct perf_stat_config *config, struct evsel *evsel, }if (config->aggr_mode ==AGGR_THREAD ) { -if (config->stats) -perf_stat__update_shadow_stats (evsel, - count->val,0 , &config->stats[thread]); -else -perf_stat__update_shadow_stats (evsel, - count->val,0 , &rt_stat); +perf_stat__update_shadow_stats (evsel, count->val, + thread, &rt_stat); }break ;case AGGR_GLOBAL :
Signature · 1 line
2.38.0.rc1.362.ged0d419d3c-goog
Namhyung Kim
maintainerPATCH 5/7 perf stat: Kill unused per-thread runtime stats
In reply to Namhyung KimNow it's using the global rt_stat, no need to use per-thread stats. Let get rid of them.
--- tools/perf/builtin-stat.c | 54 --------------------------------------- tools/perf/util/stat.h | 2 -- 2 files changed, 56 deletions(-)
tools/perf/builtin-stat.c +0 -54
@@ -292,13 +292,8 @@ static inline void diff_timespec(struct timespec *r, struct timespec *a,static void perf_stat__reset_stats (void ) { -int i; -evlist__reset_stats (evsel_list);perf_stat__reset_shadow_stats (); - -for (i =0 ; i < stat_config.stats_num; i++) -perf_stat__reset_shadow_per_stat (&stat_config.stats[i]); } static int process_synthesized_event(struct perf_tool *tool __maybe_unused,
@@ -489,46 +484,6 @@ static void read_counters(struct timespec *rs) } } -static int runtime_stat_new (struct perf_stat_config *config,int nthreads) -{ -int i; - - config->stats =calloc (nthreads,sizeof (struct runtime_stat )); -if (!config->stats) -return -1 ; - - config->stats_num = nthreads; - -for (i =0 ; i < nthreads; i++) -runtime_stat__init (&config->stats[i]); - -return 0 ; -} - -static void runtime_stat_delete (struct perf_stat_config *config) -{ -int i; - -if (!config->stats) -return ; - -for (i =0 ; i < config->stats_num; i++) -runtime_stat__exit (&config->stats[i]); - -zfree (&config->stats); -} - -static void runtime_stat_reset (struct perf_stat_config *config) -{ -int i; - -if (!config->stats) -return ; - -for (i =0 ; i < config->stats_num; i++) -perf_stat__reset_shadow_per_stat (&config->stats[i]); -} -static void process_interval (void ) {struct timespec ts, rs;
@@ -537,7 +492,6 @@ static void process_interval(void)diff_timespec (&rs, &ts, &ref_time);perf_stat__reset_shadow_per_stat (&rt_stat); -runtime_stat_reset (&stat_config);read_counters (&rs);if (STAT_RECORD ) {
@@ -1018,7 +972,6 @@ static int __run_perf_stat(int argc, const char **argv, int run_idx)evlist__copy_prev_raw_counts (evsel_list);evlist__reset_prev_raw_counts (evsel_list); -runtime_stat_reset (&stat_config);perf_stat__reset_shadow_per_stat (&rt_stat); }else {update_stats (&walltime_nsecs_stats, t1 - t0);
@@ -2514,12 +2467,6 @@ int cmd_stat(int argc, const char **argv)*/ if (stat_config.aggr_mode ==AGGR_THREAD ) {thread_map__read_comms (evsel_list->core.threads); -if (target.system_wide) { -if (runtime_stat_new (&stat_config, -perf_thread_map__nr (evsel_list->core.threads))) { - goto out; - } - } }if (stat_config.aggr_mode ==AGGR_NODE )
@@ -2660,7 +2607,6 @@ int cmd_stat(int argc, const char **argv)evlist__delete (evsel_list);metricgroup__rblist_exit (&stat_config.metric_events); -runtime_stat_delete (&stat_config);evlist__close_control (stat_config.ctl_fd, stat_config.ctl_fd_ack, &stat_config.ctl_fd_close);return status;
tools/perf/util/stat.h +0 -2
@@ -153,8 +153,6 @@ struct perf_stat_config {int run_count;int print_free_counters_hint;int print_mixed_hw_group_error; -struct runtime_stat *stats; -int stats_num;const char *csv_sep;struct stats *walltime_nsecs_stats;struct rusage ru_data;
Signature · 1 line
2.38.0.rc1.362.ged0d419d3c-goog
Namhyung Kim
maintainerPATCH 6/7 perf stat: Don't compare runtime stat for shadow stats
In reply to Namhyung KimNow it always uses the global rt_stat. Let's get rid of the field from the saved_value. When the both evsels are NULL, it'd return 0 so remove the block in the saved_value_cmp.
--- tools/perf/util/stat-shadow.c | 12 ------------ 1 file changed, 12 deletions(-)
tools/perf/util/stat-shadow.c +0 -12
@@ -35,7 +35,6 @@ struct saved_value {int ctx;int map_idx;/* cpu or thread map index */ struct cgroup *cgrp; -struct runtime_stat *stat;struct stats stats;u64 metric_total;int metric_other;
@@ -67,16 +66,6 @@ static int saved_value_cmp(struct rb_node *rb_node, const void *entry)if (a->cgrp != b->cgrp)return (char *)a->cgrp < (char *)b->cgrp ?-1 :+1 ; -if (a->evsel ==NULL && b->evsel ==NULL ) { -if (a->stat == b->stat) -return 0 ; - -if ((char *)a->stat < (char *)b->stat) -return -1 ; - -return 1 ; - } -if (a->evsel == b->evsel)return 0 ;if ((char *)a->evsel < (char *)b->evsel)
@@ -120,7 +109,6 @@ static struct saved_value *saved_value_lookup(struct evsel *evsel, .evsel = evsel, .type = type, .ctx = ctx, - .stat = st, .cgrp = cgrp, };
Signature · 1 line
2.38.0.rc1.362.ged0d419d3c-goog
Namhyung Kim
maintainerPATCH 7/7 perf stat: Rename to aggr_cpu_id.thread_idx
In reply to Namhyung KimThe aggr_cpu_id has a thread value but it's actually an index to the thread_map. To reduce possible confusion, rename it to thread_idx.
--- tools/perf/tests/topology.c | 10 +++++----- tools/perf/util/cpumap.c | 8 ++++---- tools/perf/util/cpumap.h | 2 +- tools/perf/util/stat-display.c | 12 ++++++------ 4 files changed, 16 insertions(+), 16 deletions(-)
tools/perf/tests/topology.c +5 -5
@@ -147,7 +147,7 @@ static int check_cpu_topology(char *path, struct perf_cpu_map *map)TEST_ASSERT_VAL ("Cpu map - Die ID doesn't match" , session->header.env.cpu[perf_cpu_map__cpu (map, i).cpu].die_id == id.die);TEST_ASSERT_VAL ("Cpu map - Node ID is set" , id.node ==-1 ); -TEST_ASSERT_VAL ("Cpu map - Thread is set" , id.thread ==-1 ); +TEST_ASSERT_VAL ("Cpu map - Thread IDX is set" , id.thread_idx ==-1 ); }// Test that core ID contains socket, die and core
@@ -163,7 +163,7 @@ static int check_cpu_topology(char *path, struct perf_cpu_map *map)TEST_ASSERT_VAL ("Core map - Die ID doesn't match" , session->header.env.cpu[perf_cpu_map__cpu (map, i).cpu].die_id == id.die);TEST_ASSERT_VAL ("Core map - Node ID is set" , id.node ==-1 ); -TEST_ASSERT_VAL ("Core map - Thread is set" , id.thread ==-1 ); +TEST_ASSERT_VAL ("Core map - Thread IDX is set" , id.thread_idx ==-1 ); }// Test that die ID contains socket and die
@@ -179,7 +179,7 @@ static int check_cpu_topology(char *path, struct perf_cpu_map *map)TEST_ASSERT_VAL ("Die map - Node ID is set" , id.node ==-1 );TEST_ASSERT_VAL ("Die map - Core is set" , id.core ==-1 );TEST_ASSERT_VAL ("Die map - CPU is set" , id.cpu.cpu ==-1 ); -TEST_ASSERT_VAL ("Die map - Thread is set" , id.thread ==-1 ); +TEST_ASSERT_VAL ("Die map - Thread IDX is set" , id.thread_idx ==-1 ); }// Test that socket ID contains only socket
@@ -193,7 +193,7 @@ static int check_cpu_topology(char *path, struct perf_cpu_map *map)TEST_ASSERT_VAL ("Socket map - Die ID is set" , id.die ==-1 );TEST_ASSERT_VAL ("Socket map - Core is set" , id.core ==-1 );TEST_ASSERT_VAL ("Socket map - CPU is set" , id.cpu.cpu ==-1 ); -TEST_ASSERT_VAL ("Socket map - Thread is set" , id.thread ==-1 ); +TEST_ASSERT_VAL ("Socket map - Thread IDX is set" , id.thread_idx ==-1 ); }// Test that node ID contains only node
@@ -205,7 +205,7 @@ static int check_cpu_topology(char *path, struct perf_cpu_map *map)TEST_ASSERT_VAL ("Node map - Die ID is set" , id.die ==-1 );TEST_ASSERT_VAL ("Node map - Core is set" , id.core ==-1 );TEST_ASSERT_VAL ("Node map - CPU is set" , id.cpu.cpu ==-1 ); -TEST_ASSERT_VAL ("Node map - Thread is set" , id.thread ==-1 ); +TEST_ASSERT_VAL ("Node map - Thread IDX is set" , id.thread_idx ==-1 ); }perf_session__delete (session);
tools/perf/util/cpumap.c +4 -4
@@ -229,7 +229,7 @@ static int aggr_cpu_id__cmp(const void *a_pointer, const void *b_pointer)else if (a->core != b->core)return a->core - b->core;else -return a->thread - b->thread; +return a->thread_idx - b->thread_idx; } struct cpu_aggr_map *cpu_aggr_map__new(const struct perf_cpu_map *cpus,
@@ -667,7 +667,7 @@ const struct perf_cpu_map *cpu_map__online(void) /* thread unsafe */bool aggr_cpu_id__equal (const struct aggr_cpu_id *a,const struct aggr_cpu_id *b) { -return a->thread == b->thread && +return a->thread_idx == b->thread_idx && a->node == b->node && a->socket == b->socket && a->die == b->die &&
@@ -677,7 +677,7 @@ bool aggr_cpu_id__equal(const struct aggr_cpu_id *a, const struct aggr_cpu_id *bbool aggr_cpu_id__is_empty (const struct aggr_cpu_id *a) { -return a->thread ==-1 && +return a->thread_idx ==-1 && a->node ==-1 && a->socket ==-1 && a->die == -1 &&
@@ -688,7 +688,7 @@ bool aggr_cpu_id__is_empty(const struct aggr_cpu_id *a)struct aggr_cpu_id aggr_cpu_id__empty (void ) {struct aggr_cpu_id ret = { - .thread =-1 , + .thread_idx =-1 , .node =-1 , .socket =-1 , .die =-1 ,
tools/perf/util/cpumap.h +1 -1
@@ -10,7 +10,7 @@/** Identify where counts are aggregated, -1 implies not to aggregate. */ struct aggr_cpu_id {/** A value in the range 0 to number of threads. */ -int thread; +int thread_idx;/** The numa node X as read from /sys/devices/system/node/nodeX. */ int node;/**
tools/perf/util/stat-display.c +6 -6
@@ -189,14 +189,14 @@ static void aggr_printout(struct perf_stat_config *config,case AGGR_THREAD :if (config->json_output) {fprintf (config->output,"\"thread\" : \"%s-%d\", " , -perf_thread_map__comm (evsel->core.threads, id.thread), -perf_thread_map__pid (evsel->core.threads, id.thread)); +perf_thread_map__comm (evsel->core.threads, id.thread_idx), +perf_thread_map__pid (evsel->core.threads, id.thread_idx)); }else {fprintf (config->output,"%*s-%*d%s" , config->csv_output ?0 :16 , -perf_thread_map__comm (evsel->core.threads, id.thread), +perf_thread_map__comm (evsel->core.threads, id.thread_idx), config->csv_output ?0 :-8 , -perf_thread_map__pid (evsel->core.threads, id.thread), +perf_thread_map__pid (evsel->core.threads, id.thread_idx), config->csv_sep); }break ;
@@ -453,7 +453,7 @@ static int first_shadow_map_idx(struct perf_stat_config *config,return perf_cpu_map__idx (cpus, id->cpu);if (config->aggr_mode ==AGGR_THREAD ) -return id->thread; +return id->thread_idx;if (!config->aggr_get_id)return 0 ;
@@ -946,7 +946,7 @@ static struct perf_aggr_thread_value *sort_aggr_thread( buf[i].counter = counter; buf[i].id =aggr_cpu_id__empty (); - buf[i].id.thread = thread; + buf[i].id.thread_idx = thread; buf[i].uval = uval; buf[i].val = val; buf[i].run = run;
Signature · 1 line
2.38.0.rc1.362.ged0d419d3c-goog
Arnaldo Carvalho de Melo
maintainerEm Fri, Sep 30, 2022 at 01:21:03PM -0700, Namhyung Kim escreveu:
Quoted 6 lines
Hello,
I'm working on perf stat and I found some items to clean up. This time I removed runtime stats for per-thread aggregation mode which we can simply use thread map index to compare the shadow stat values in the
rt_stat.
Quoted 8 lines
Changes from v1)
* renaming missing parts in headers (James)
* add a comment formap_idx(Ian)
* add Reviewed-by from JamesThe code is available at 'perf/stat-cleanup-v2' branch in
git://git.kernel.org/pub/scm/linux/kernel/git/namhyung/linux-perf.git
Applied locally, testing.
- Arnaldo
Quoted 27 lines
Thanks,
NamhyungNamhyung Kim (7):
perf stat: Convert perf_stat_evsel.res_stats array perf stat: Don't callperf_stat_evsel_id_init()repeatedly perf stat: Renamesaved_value->cpu_map_idx
perf stat: Use thread map index for shadow stat perf stat: Kill unused per-thread runtime stats perf stat: Don't compare runtime stat for shadow stats perf stat: Rename to aggr_cpu_id.thread_idxtools/perf/builtin-stat.c | 54 ------ tools/perf/tests/topology.c | 10 +- tools/perf/util/cpumap.c | 8 +- tools/perf/util/cpumap.h | 2 +- tools/perf/util/stat-display.c | 32 ++-- tools/perf/util/stat-shadow.c | 320 ++++++++++++++++----------------- tools/perf/util/stat.c | 20 +-- tools/perf/util/stat.h | 10 +- 8 files changed, 189 insertions(+), 267 deletions(-)base-commit:
62e64c9d2fd12839c02f1b3e8b873e7cb34e8720Signature · 1 line
2.38.0.rc1.362.ged0d419d3c-goog
Signature · 1 line
- Arnaldo